Digital imaging method and system for ground penetrating radar road defect detection considering data asymmetry

Abstract

The present invention discloses a digital imaging method and system for ground-penetrating radar (GPR) road defect detection that takes data asymmetry into account. The method first collects road electromagnetic echo signal data to generate two-dimensional B-Scan GPR echo data (range-two-way travel time). The B-Scan data is preprocessed for DC removal and background removal. An adaptive calculation method is used to determine the optimal imaging threshold, ensuring that the image background remains stable at a grayscale value of approximately 128, thereby enhancing defect information. Finally, the B-Scan data values ​​are converted to grayscale values ​​between 0 and 255 using a grayscale linear conversion method to generate a GPR road grayscale image. The method comprehensively considers the impact of GPR data asymmetry on the imaging effects of background and defect information, enhancing defect information while ensuring image background quality. This overcomes the inability of existing methods to simultaneously ensure both background quality and defect information enhancement. The method is suitable for various road inspections and provides high-quality GPR inspection images for inspectors or machine learning models.

Description

Technical Field

[0001] The present invention belongs to the technical field of ground penetrating radar digital imaging, and mainly relates to a ground penetrating radar road defect detection digital imaging method and system taking into account data asymmetry. Background Art

[0002] With the sustained development of my country's national economy and the accelerating pace of urbanization, roads have become a vital means of public transportation and land transport. Their health directly impacts the safety of public transportation and the smooth and secure flow of land transport. In recent years, the explosive growth of vehicle ownership and the rapid development of infrastructure, coupled with adverse weather conditions, have increased the risk of road defects such as cracks, voids, and uneven settlement. If these defects are not promptly detected and repaired, they will worsen over time and may even cause major accidents, resulting in traffic disruptions, property damage, and casualties. Therefore, to prevent these major incidents, effective road defect detection technology is urgently needed.

[0003] Ground-penetrating radar (GPR) technology is a novel, visual, nondestructive road inspection method that can accurately detect hidden road defects with high resolution, high efficiency, and real-time imaging. This technology uses a transmitting antenna to emit high-frequency electromagnetic waves into the ground. When these waves encounter interfaces with electrical differences, they are reflected, and a receiving antenna picks up the reflected electromagnetic waves as an A-Scan signal. As the inspection vehicle continuously travels along the survey line, the receiving antenna picks up echo signals at different locations, generating a series of A-Scan signals. These signals are then converted into two-dimensional B-Scan GPR echo data (range-two-way travel time). The B-Scan signals are then converted into images. By analyzing the characteristics of GPR images, the distribution of the subsurface can be effectively analyzed. However, in GPR systems, defects located deep underground have weak reflected signals, which can be buried in clutter signals such as the more energetic inter-antenna coupling signal and reflections from the road structure, significantly interfering with defect detection and identification.

[0004] In ground-penetrating radar images, most pixels are background information, while defect information is the information of interest. Therefore, the poor quality and instability of the background and the low contrast of the defect information will reduce the interpretability of the image, making it difficult to provide high-quality ground-penetrating radar images for inspectors or machine learning models, which can easily lead to misjudgments and omissions in road defect detection.

[0005] There are two main traditional GPR digital imaging methods. One method is based on linear conversion of quantization bits. This involves linearly converting the data value of each pixel into a grayscale value based on the range of quantization bits acquired by the hardware, thereby generating a GPR grayscale image. However, while this method effectively removes background information related to the dielectric layer after preprocessing, the data preprocessing algorithm reduces the overall value of the data matrix. While linear conversion to image grayscale based on the range of quantization bits effectively preserves the image background quality, the grayscale value of defect information approaches the background grayscale value, reducing the contrast of the defect information in the image. Another method is linear conversion based on the data range. This method first dynamically calculates the minimum and maximum values ​​in the B-scan data matrix, and then linearly converts the data value of each pixel into a grayscale value based on the range of the maximum values. This method primarily aims to maximize the enhancement of defect information. Since defect information in a B-scan is often located near the maximum value of the entire digital matrix, this imaging method can keep the grayscale value of the defect information as close to 0 or 255 as possible, thereby enhancing the defect information in the image. However, in actual underground structures, the dielectric layers fluctuate to a certain degree. Coupled with the influence of numerous noises in the underground environment, even after preprocessing, a significant amount of interference information remains. This interference data causes the minimum and maximum values ​​of the entire B-scan data to be significantly asymmetric about zero. In this case, using this imaging method can cause the background to appear dark or bright, resulting in poor background stability in the image, which can affect both manual judgment and recognition results of machine learning models.

[0006] Therefore, existing research has not involved imaging methods that take into account the asymmetric characteristics of GPR data, especially the lack of research on GPR data imaging methods with stable background quality, good defect enhancement effect, and adaptability to various road inspections. Summary of the Invention

[0007] The present invention addresses the problem that existing technologies fail to account for the asymmetry of GPR data, resulting in unstable background quality and poor defect enhancement. This invention provides a digital imaging method and system for GPR road defect detection that takes data asymmetry into account. The method first acquires road electromagnetic echo signal data to generate two-dimensional B-Scan GPR echo data, representing range and two-way travel time. The B-Scan data is then preprocessed for DC removal and background removal. An adaptive calculation method is used to determine the optimal imaging threshold, ensuring that the image background remains stable around a grayscale value of 128, thereby enhancing defect information. Finally, the B-Scan data values ​​are converted to grayscale values ​​between 0 and 255 using a grayscale linear conversion method to generate a GPR road grayscale image. This method comprehensively considers the impact of GPR data asymmetry on the imaging of both background and defect information. While maintaining image background quality, it also enhances defect information. This method is adaptable to various road inspection environments, overcoming the inefficiency and adaptability of existing methods. It provides inspectors or machine learning models with high-quality GPR images with stable background quality and high contrast, thereby improving the detection and recognition rates of underground road defects.

[0008] In order to achieve the above object, the technical solution adopted by the present invention is: a digital imaging method for ground penetrating radar road defect detection considering data asymmetry, comprising the following steps:

[0009] S1, data acquisition: collecting road electromagnetic echo signal data, which reflects the underground medium condition of the road and forms two-dimensional B-Scan ground penetrating radar echo data as distance-two-way travel time;

[0010] S2, data processing: processing the two-dimensional B-Scan ground penetrating radar echo data collected in step S1, wherein the processing includes at least DC removal preprocessing and background removal preprocessing.

[0011] S3, optimal threshold calculation: using adaptive calculation method to determine the optimal imaging threshold, ensuring that the image background is stable at a grayscale value of around 128, and enhancing defect information;

[0012] S4, image conversion: convert the B-Scan data value into a grayscale value [0, 255] through grayscale linear conversion to generate a ground penetrating radar road grayscale image.

[0013] As an improvement of the present invention, in step S1, the road electromagnetic wave signal is collected by the ground penetrating radar system on the detection vehicle, and after AD sampling and conversion, it is converted into a digital signal to form a signal in the form of A-Scan. The detection vehicle moves along the survey line, and the receiving antenna receives the echo signal at different positions to form a two-dimensional B-Scan ground penetrating radar echo data of distance-two-way travel time.

[0014] As an improvement of the present invention, the DC removal preprocessing in step S2 is specifically as follows: each A-Scan data is processed in sequence. For the j-th A-Scan, the average value of the echo data value of the current channel is first calculated, and the average value is used as the DC offset of the current channel. Then, the average value is subtracted from all data values ​​of the channel, that is:

[0015]

[0016] Where i and j are the number of rows and columns of the B-Scan data being processed, respectively; is the data after DC preprocessing; X(i,j) is the original data of the ground penetrating radar; N is the number of sampling points of each data, that is, the total number of rows of B-Scan.

[0017] As another improvement of the present invention, the background removal preprocessing in step S2 is specifically as follows: first, the column average of all columns of the entire B-Scan data matrix is ​​calculated, and then the column average is subtracted from each A-Scan data value to obtain the background-removed B-Scan data, that is:

[0018]

[0019] Where i and j are the number of rows and columns of the B-Scan data being processed, respectively; M is the total number of columns of the B-Scan data; is the data before background removal preprocessing; X'(i,j) represents the data after background removal preprocessing.

[0020] As another improvement of the present invention, the adaptive calculation method of step S3 is specifically as follows:

[0021] S31: Calculate the maximum and minimum values ​​of the B-Scan data, traverse the entire B-Scan data, and calculate the minimum value min and the maximum value max of all data values. The relationship satisfies the following formula:

[0022] min≤X'(i,j)≤max

[0023] S32: Calculating the optimal threshold for imaging, the step further comprising:

[0024] S321: Take the absolute values ​​of the minimum value min and the maximum value max respectively, and take the maximum value of the two as the optimal threshold t for grayscale linear conversion;

[0025]

[0026] S322: Determine two optimal thresholds t1 and t2 for grayscale conversion. The two thresholds t1 and t2 correspond to the values ​​of 0 and 255 for linear conversion to grayscale values, respectively. The calculation formula is as follows:

[0027] For the optimal threshold t1 of linear grayscale value 0 conversion, the specific operations are:

[0028] t1=-t

[0029] For the optimal threshold t2 for linear grayscale value 255 conversion, the specific operations are:

[0030] t2=t.

[0031] As another improvement of the present invention, in the grayscale linear conversion of step S4, the grayscale value G(i, j) of each pixel is calculated as follows:

[0032]

[0033] In order to achieve the above-mentioned purpose, the present invention also adopts the following technical solution: a digital imaging system for ground-penetrating radar road defect detection taking into account data asymmetry, comprising a computer program, characterized in that when the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] (1) The method proposed in the present invention takes into account the impact of B-Scan data asymmetry on imaging quality. Under the premise of ensuring stable background quality, the defect information is enhanced, thereby improving the quality of the ground penetrating radar image. It can provide ground penetrating radar images with stable background quality and high defect information contrast for inspection personnel or machine learning models, overcoming the shortcomings of existing methods that cannot simultaneously ensure background quality and defect information enhancement.

[0036] (2) The method proposed in the present invention adaptively calculates the optimal imaging threshold based on the collected B-Scan data, and can be applied to road defect detection in various road inspection scenarios;

[0037] (3) The method proposed in the present invention improves the quality of ground penetrating radar images during the digital imaging process, avoids the need to use image processing methods to improve the image quality subsequently, and improves the overall detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flow chart of the steps of the method of the present invention;

[0039] Figure 2 3 is a flowchart of the steps of adaptively calculating the optimal imaging threshold in step S3 of the method of the present invention;

[0040] Figure 3 is a data histogram after data preprocessing in Example 2 of the present invention;

[0041] Figure 4is a ground penetrating radar image generated by Example 2 of the present invention;

[0042] Figure 5 is the grayscale histogram of the ground penetrating radar image in Example 2 of the present invention;

[0043] Figure 6 This is a grayscale value curve diagram of the channel where the minimum value of the data is located after data preprocessing in Example 2 of the present invention. DETAILED DESCRIPTION

[0044] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention.

[0045] Example 1

[0046] Digital imaging method for ground penetrating radar road defect detection considering data asymmetry, such as Figure 1 As shown, the following steps are included:

[0047] Step S1: Collect the original echo data of the ground penetrating radar of the road B-Scan

[0048] To achieve effective visual nondestructive testing of roads, a ground-penetrating radar (GPR) system is installed on an inspection vehicle. The system consists of a main control unit (CCU), a transmitter, a receiver, a transmitting antenna, and a receiving antenna. The CCU controls the transmitter and receiver by issuing a series of control commands. The transmitter transmits electromagnetic waves into the ground via the transmitting antenna, while the receiver collects the reflected electromagnetic wave signals via the receiving antenna. After analog-to-digital conversion, the electromagnetic wave signals, which contain information about the underground medium, are converted into digital signals, forming an A-Scan signal. As the inspection vehicle moves along the survey line, the receiving antenna picks up echo signals at different locations, generating a series of A-Scan signals that form two-dimensional B-Scan GPR echo data, representing distance and two-way travel time.

[0049] Step S2: Processing the ground penetrating radar B-Scan data

[0050] The DC drift of the GPR system and the road structure layer information significantly interfere with the underground defect reflection signal. Therefore, it is necessary to preprocess the collected raw B-Scan data to eliminate interference information and highlight the underground defect information, thereby better interpreting the data anomalies. This includes the following two sub-steps:

[0051] Sub-step S21: DC removal pre-processing

[0052] To effectively remove the system DC drift information in the ground penetrating radar B-Scan data, the B-Scan data is subjected to DC removal processing, and each A-Scan is processed in turn. For the j-th A-Scan, the average value of the echo data value of the current channel is first calculated. This average value is used as the DC drift of the current channel. Then, this average value is subtracted from all data values ​​of the channel. The specific formula is as follows:

[0053]

[0054] Where i and j are the number of rows and columns of the B-Scan data being processed, respectively. is the data after DC preprocessing, X(i,j) is the original data of GPR, N is the number of sampling points of each data, that is, the total number of rows of B-Scan;

[0055] Sub-step S22: Background removal preprocessing

[0056] In order to effectively remove the road structure dielectric layer information in the ground penetrating radar B-Scan, the B-Scan data after DC removal is subjected to background removal. Common background removal methods include mean background removal and sliding window background removal. In order to improve the efficiency of preprocessing, the present invention adopts the mean method for background removal. The direct wave signal has the characteristics of isochronism and stability. The direct wave with large signal amplitude in the B-Scan data is removed. The column average of all columns of the entire B-Scan data matrix is ​​first calculated. Then, the column average is subtracted from each A-Scan data value to obtain the B-Scan data after background removal. The specific formula is as follows:

[0057]

[0058] Where i and j are the number of rows and columns of the B-Scan data being processed, M is the total number of columns of the B-Scan data, is the data before background removal preprocessing, and X'(i,j) represents the data after background removal preprocessing.

[0059] The DC removal and background removal data preprocessing methods can remove clutter information and enhance the defect information of the image, thereby improving the interpretability of the defect data.

[0060] Step S3: Adaptively calculate the optimal imaging threshold considering data asymmetry

[0061] In order to obtain a ground penetrating radar image with stable background quality and high contrast, the present invention designs a method for calculating the optimal threshold of adaptive B-Scan digital imaging from the perspective of data asymmetry. The algorithm flow chart is as follows: Figure 2This algorithm can effectively ensure that the background of the ground penetrating radar image is stable around the grayscale value of 128, while enhancing the defect information. It specifically includes the following two sub-steps: Figure 2 As shown: Sub-step S31: Calculate the maximum and minimum values ​​of B-Scan data

[0062] First, traverse the entire B-Scan data and calculate the minimum value min and maximum value max of all data values. The relationship satisfies the following formula:

[0063] min≤X'(i,j)≤max

[0064] Sub-step S32: Calculate the optimal threshold for imaging

[0065] To effectively address the difficulty of simultaneously maintaining high image background quality and high contrast for defect information due to the asymmetry of ground-penetrating radar B-Scan data, an algorithm for adaptively calculating the optimal imaging threshold was designed based on the fact that background data values ​​in preprocessed B-Scan data are primarily concentrated near 0, while defect information values ​​are relatively small or large. This algorithm maintains a stable image background while enhancing defect information and is applicable to various road inspections. The calculation process is as follows:

[0066] In sub-step S321, the absolute values ​​of the minimum value min and the maximum value max are taken respectively, and the maximum value of the two is taken as the optimal threshold t for grayscale linear conversion. The specific calculation formula is as follows:

[0067]

[0068] Sub-step S322 determines two optimal thresholds t1 and t2 for grayscale conversion. The two thresholds t1 and t2 correspond to the values ​​linearly converted to grayscale values ​​of 0 and 255, respectively. The calculation formula is as follows:

[0069] For the optimal threshold t1 of linear grayscale value 0 conversion, the specific operations are:

[0070] t1=-t

[0071] For the optimal threshold t2 for linear grayscale value 255 conversion, the specific operations are:

[0072] t2=t

[0073] Step S4: Converting the ground penetrating radar B-Scan data into a grayscale image

[0074] According to the calculated optimal thresholds t1 and t2, the B-Scan data values ​​are converted into grayscale values ​​[0, 255] through grayscale linear conversion to generate a ground penetrating radar road grayscale image. The grayscale value G(i, j) of each pixel is calculated as follows:

[0075]

[0076] Example 2

[0077] The above method is applied to a specific example below to demonstrate its technical effect.

[0078] The digital imaging method for detecting road defects using ground penetrating radar considering data asymmetry includes the following steps:

[0079] S1, data acquisition: collecting A-scan electromagnetic echo signal data and obtaining B-scan ground penetrating radar data with a dimension of 512×512. The signal reflects the underground medium condition of the road;

[0080] S2, data processing: Process the two-dimensional B-Scan ground penetrating radar echo data collected in step S1 to remove the system DC drift information and dielectric layer information. The processing includes at least DC removal preprocessing and background removal preprocessing. The data is processed by removing DC and averaging methods to remove background. The data histogram after data preprocessing is as follows: Figure 3 As shown, Figure 3 It shows that most of the data values ​​are background information, mainly distributed around 0, with a maximum value of 15644 and a minimum value of -25930. There is a clear asymmetry between the maximum and minimum values.

[0081] S3, optimal threshold calculation: using an adaptive calculation method to determine the optimal imaging threshold, ensuring that the image background is stable at a grayscale value of around 128 and enhancing defect information. In this embodiment, the optimal imaging thresholds t1 and t2 calculated by the adaptive algorithm are -25930 and 25930, respectively.

[0082] S4, image conversion: by grayscale linear conversion, the B-Scan data value is converted into grayscale value [0,255] to generate the ground penetrating radar road grayscale image, such as Figure 4 shown.

[0083] Depend on Figure 4 It can be seen that the overall background of the image is gray and the defect information is obvious. Figure 5 The grayscale histogram of the ground penetrating radar image generated by the embodiment of the present invention is Figure 5 It can be seen that the grayscale values ​​of most pixels in the image are around 128, that is, the background is gray, which effectively ensures the background quality of the GPR image. Figure 6 This is a grayscale value curve diagram of the channel where the data minimum value is located after data preprocessing in the embodiment provided by the present invention. The grayscale values ​​of most data points in this channel are around 128, among which the grayscale value corresponding to the data minimum value is 0, and the defect information is fully enhanced on the image.

[0084] In summary, this patent fully considers the asymmetry of data, achieves the purpose of stabilizing the background quality of the image after ground penetrating radar imaging, and at the same time enhances the defect information. Figure 3 It can be seen that after data preprocessing, the B-Scan data has obvious asymmetry. Figure 4 It is clear that the background of the image after imaging is neither too dark nor too bright, demonstrating good and stable background quality. Defect information is also enhanced, with clear outlines. Therefore, the patented method can effectively ensure stable background quality and enhanced defect information after ground-penetrating radar data imaging. Figure 5 After verifying that this method is used, the image background value is stable around 128, which proves that the image background quality is good. Because the grayscale value ranges from 0 to 255, where 0 is black and 255 is white, the closer the background is to 128, the grayer it appears. The closer the defect information is to 0 and 255, the higher the contrast of the defect information, and the more obvious the defect information is. Therefore, Figure 6 The grayscale value of the defect information is close to 0, which proves that the defect information is enhanced after imaging.

[0085] It should be noted that the above content merely illustrates the technical idea of ​​the present invention and cannot be used to limit the scope of protection of the present invention. For ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications all fall within the scope of protection of the claims of the present invention.

Claims

1. A digital imaging method for ground penetrating radar road defect detection considering data asymmetry, characterized in that: The following steps are involved: S1, data acquisition: collecting road electromagnetic echo signal data, which reflects the underground medium condition of the road and forms two-dimensional B-Scan ground penetrating radar echo data as distance-two-way travel time; S2, data processing: processing the two-dimensional B-Scan ground penetrating radar echo data collected in step S1, wherein the processing includes at least DC removal preprocessing and background removal preprocessing. S3, optimal threshold calculation: Use an adaptive calculation method to determine the optimal imaging threshold, ensure that the image background is stable at a grayscale value of around 128, and enhance defect information; the adaptive calculation method is specifically as follows: S31: Calculate the maximum and minimum values ​​of the B-Scan data, traverse the entire B-Scan data, and calculate the minimum value min and the maximum value max of all data values. The relationship satisfies the following formula: min≤X'(i,j)≤max S32: Calculating the optimal threshold for imaging, the step further comprising: S321: Take the absolute values ​​of the minimum value min and the maximum value max respectively, and take the maximum value of the two as the optimal threshold t for grayscale linear conversion; S322: Determine two optimal thresholds t1 and t2 for grayscale conversion. The two thresholds t1 and t2 correspond to the values ​​of 0 and 255 for linear conversion to grayscale values, respectively. The calculation formula is as follows: For the optimal threshold t1 of linear grayscale value 0 conversion, the specific operations are: t1=-t For the optimal threshold t2 for linear grayscale value 255 conversion, the specific operations are: t2 = t; S4, image conversion: convert the B-Scan data value into a grayscale value [0, 255] through grayscale linear conversion to generate a ground penetrating radar road grayscale image.

2. The digital imaging method for ground penetrating radar road defect detection considering data asymmetry according to claim 1, characterized in that: In step S1, the electromagnetic wave signal of the road is collected by the ground penetrating radar system on the detection vehicle, and after AD sampling and conversion, it is converted into a digital signal to form an A-Scan signal. The detection vehicle moves along the measurement line, and the receiving antenna receives the echo signal at different positions to form a two-dimensional B-Scan ground penetrating radar echo data of distance-two-way travel time.

3. The digital imaging method for ground penetrating radar road defect detection considering data asymmetry according to claim 2, characterized in that: The DC removal preprocessing in step S2 is specifically as follows: each A-Scan data is processed in turn. For the j-th A-Scan, the average value of the echo data value of the current channel is first calculated, and the average value is used as the DC offset of the current channel. Then, the average value is subtracted from all data values ​​of the channel, that is: Where i and j are the number of rows and columns of the B-Scan data being processed, respectively; is the data after DC preprocessing; X(i,j) is the original data of the ground penetrating radar; N is the number of sampling points of each data, that is, the total number of rows of B-Scan.

4. The digital imaging method for ground penetrating radar road defect detection considering data asymmetry according to claim 2, characterized in that: The background removal preprocessing in step S2 is specifically as follows: first, the column average of all columns of the entire B-Scan data matrix is ​​calculated, and then the column average is subtracted from each A-Scan data value to obtain the background-removed B-Scan data, that is: Where i and j are the number of rows and columns of the B-Scan data being processed, respectively; M is the total number of columns of the B-Scan data; is the data before background removal preprocessing; X'(i,j) represents the data after background removal preprocessing.

5. The digital imaging method for ground penetrating radar road defect detection considering data asymmetry according to claim 4, characterized in that: In the grayscale linear conversion of step S4, the grayscale value G(i,j) of each pixel is calculated as follows:

6. A digital imaging system for ground penetrating radar road defect detection taking into account data asymmetry, including a computer program, characterized by: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images